readability support

Author: web64

src/NlpClient.php

@@ -70,6 +70,26 @@ public function newspaperUrl( $url )
 		return ( !empty($data['newspaper']) ) ? $data['newspaper'] : null;
 	}
 
+	
+	/**
+	 * 	Readability Article Extraction from URL
+	 */
+	public function readabilityUrl( $url )
+	{
+		$data = $this->get_call('/readability', ['url' => $url ] );
+
+		return ( !empty($data['data']) ) ? $data['data'] : null;
+	}
+
+	/**
+	 * 	Readability Article Extraction from HTML
+	 */
+	public function readabilityHTML( $html )
+	{
+		$data = $this->post_call('/readability', ['html' => $html ] );
+
+		return ( !empty($data['data']) ) ? $data['data'] : null;
+	}
 
 	/**
 	 * 		Get neighbouring words

tests/Unit/CoreNlpTest.php

@@ -10,7 +10,7 @@ class CoreNlpTest extends TestCase
     public function test_core_nlp()
     {
         $corenlp = new \Web64\Nlp\CoreNlp();
-        echo PHP_EOL. PHP_EOL;
+        //echo PHP_EOL. PHP_EOL;
         $text = "Catalonia: Ex-police chief Trapero charged with sedition. The former chief of Catalonia's police force, Josep Lluis Trapero, has been charged over events linked with last year's independence referendum.";
 
         $text = "German Foreign Ministry investigating arrest of German reporter in Turkey.

tests/Unit/NewspaperTest.php

@@ -6,26 +6,26 @@
 
 class NewspaperTest extends TestCase 
 {
-    // /** @test */
-    // public function url_article_extraction()
-    // {
-    //     $nlp = new \Web64\Nlp\NlpClient( $this->nlpserver_config['hosts'], $this->nlpserver_config['debug'] );
+    /** @test */
+    public function url_article_extraction()
+    {
+        $nlp = new \Web64\Nlp\NlpClient( $this->nlpserver_config['hosts'], $this->nlpserver_config['debug'] );
 
-    //     $newspaper = $nlp->newspaperUrl('http://www.bbc.com/news/science-environment-43710766');
+        $newspaper = $nlp->newspaperUrl('https://github.com/web64/nlpserver');
 
-    //     $this->msg( $newspaper );
-    //     $this->assertNotEmpty($newspaper);
-    // }
+        $this->msg( $newspaper );
+        $this->assertNotEmpty($newspaper);
+    }
 
     /** @test */
     public function html_article_extraction()
     {
         $nlp = new \Web64\Nlp\NlpClient( $this->nlpserver_config['hosts'], $this->nlpserver_config['debug'] );
 
-        $html = file_get_contents( 'http://www.bbc.com/news/science-environment-43710766' );
+        $html = file_get_contents( 'https://github.com/web64/nlpserver' );
         $newspaper = $nlp->newspaperHtml( $html );
 
-        //$this->msg( $newspaper );
+        $this->msg( $newspaper );
 
         $this->assertNotEmpty($newspaper);
         $this->assertNotEmpty($newspaper['title']);

tests/Unit/PolyglotTest.php

@@ -18,7 +18,7 @@ public function entity_extraction()
 
         $polyglot = $nlp->polyglot( $text, 'en' );
 
-        //$this->msg( $polyglot );
+        $this->msg( $polyglot );
 
         $this->assertNotEmpty( $polyglot->data );
         $this->assertArrayHasKey('sentiment', $polyglot->data, "Missing sentiment");

tests/Unit/ReadabilityTest.php

@@ -0,0 +1,37 @@
+<?php
+
+namespace Tests\Unit;
+
+use Tests\TestCase;
+
+class ReadabilityTest extends TestCase 
+{
+    /** @test */
+    public function readability_url_article_extraction()
+    {
+        $nlp = new \Web64\Nlp\NlpClient( $this->nlpserver_config['hosts'], $this->nlpserver_config['debug'] );
+
+        $article = $nlp->readabilityUrl('https://github.com/web64/nlpserver');
+
+        
+        //$this->msg( $article );
+        $this->assertNotEmpty($article);
+    }
+
+    /** @test */
+    public function readability_html_article_extraction()
+    {
+        $nlp = new \Web64\Nlp\NlpClient( $this->nlpserver_config['hosts'], $this->nlpserver_config['debug'] );
+
+        $html = file_get_contents( 'https://github.com/web64/nlpserver' );
+        $article = $nlp->readabilityHtml( $html );
+        
+        //$this->msg( $article );
+
+        $this->assertNotEmpty($article);
+        $this->assertNotEmpty($article['title']);
+        $this->assertNotEmpty($article['short_title']);
+        $this->assertNotEmpty($article['content']);
+        $this->assertNotEmpty($article['summary']);
+    }
+}